New rig, high temperatures
-
Thanks for the reply, Kevlar and crazyinside.
To answer the last question first, the fans are on auto and they are working hard! They are running between 3400 - 4200 RPM depending on temperature.
I installed MSI Afterburner even though the 7970 is from XFX and the 7990 is from Asus. It should work for adjusting the voltage regardless, right? Or do I maybe need to do some tricky warranty voiding flashing? :-\
I set the “Power Limit (%)” in afterburner to -20, but it didn’t seem to make any difference. But I noticed that the HW errors are racking up pretty fast. From what I understand, this should only happen when you push the intensity, but I’m still only at intensity 13.
If I go for intensity 12, the HW problems don’t show up as fast, but hash rate drops a lot, which is kinda opposite of what I want.This is what my settings batch file looks like: cgminer.exe --scrypt -o stratum+tcp://pool.d2.cc:3333 -u X -p X --intensity 13 --shaders 2048
Am I doing something wrong there? Or maybe it’s a driver issue? As mentioned, I am using the latest driver from AMD. I have Catalyst Control Center installed. Is it useful for anything or should I remove it?
Again, thanks for helping out! :)
-
where are these fans positioned? they should be in front of the cards or slightly angled towards the fans. The side or back is a waste. if you have them in back of the cards it will actually cause the air to push against the cards exhaust. ( most cards take air from the case and push it out the back )
-
[quote author=HizzleFizzle link=topic=5398.msg39811#msg39811 date=1386208504]
I installed MSI Afterburner even though the 7970 is from XFX and the 7990 is from Asus. It should work for adjusting the voltage regardless, right? Or do I maybe need to do some tricky warranty voiding flashing? :-\
[b]Afterburner will either work or it wont…i have had to use sapphire trixx instead because one will work and the other wont[/b]I set the “Power Limit (%)” in afterburner to -20, but it didn’t seem to make any difference. But I noticed that the HW errors are racking up pretty fast. From what I understand, this should only happen when you push the intensity, but I’m still only at intensity 13.
If I go for intensity 12, the HW problems don’t show up as fast, but hash rate drops a lot, which is kinda opposite of what I want.
[b]7970s will only run on I13 well, anything else will not be optimal, the HW errors will go away with proper settings[/b]This is what my settings batch file looks like: cgminer.exe --scrypt -o stratum+tcp://pool.d2.cc:3333 -u X -p X --intensity 13 --shaders 2048
[b]These are not all of your settings. Cgminer has chosen a TC automatically for you and that could be bad…i have seen it before. Look for a file similar in name to this:
scrypt130302Tahitiglg2tc21105w256l4.bin
and post what your tc value is in the name. Mine would be a tc of 21105. Also read this and use the template to the best of your knowledge:
[url=http://forum.feathercoin.com/index.php/topic,5418.0.html]http://forum.feathercoin.com/index.php/topic,5418.0.html[/url][/b] -
[quote name=“HizzleFizzle” post=“39811” timestamp=“1386208504”]
Thanks for the reply, Kevlar and crazyinside.To answer the last question first, the fans are on auto and they are working hard! They are running between 3400 - 4200 RPM depending on temperature.
I installed MSI Afterburner even though the 7970 is from XFX and the 7990 is from Asus. It should work for adjusting the voltage regardless, right? Or do I maybe need to do some tricky warranty voiding flashing? :-\
I set the “Power Limit (%)” in afterburner to -20, but it didn’t seem to make any difference. But I noticed that the HW errors are racking up pretty fast. From what I understand, this should only happen when you push the intensity, but I’m still only at intensity 13.
If I go for intensity 12, the HW problems don’t show up as fast, but hash rate drops a lot, which is kinda opposite of what I want.This is what my settings batch file looks like: cgminer.exe --scrypt -o stratum+tcp://pool.d2.cc:3333 -u X -p X --intensity 13 --shaders 2048
Am I doing something wrong there? Or maybe it’s a driver issue? As mentioned, I am using the latest driver from AMD. I have Catalyst Control Center installed. Is it useful for anything or should I remove it?
Again, thanks for helping out! :)
[/quote]If you are getting hw errors then your thread concurrency is wrong. Add this to your batch file. “–thread-concurrency 8192”
-
I would take it out of the plastic box and see what your temps do. sounds like your box is creating a pool of hot air to sit where a case as flow through design. Just because there are holes does not mean it will cool better. I would also try a box fan or something blowing down into the box, basically something to force cold air in.
-
Can you show us a picture of your rig?
What is the ambient temperature where you have the rig? (and is it summer there/do you live in a desert etc :) )
-
XFX has really sucky fan.
And 7990 cards are notoriously hot (dual 7970 on a single pcb…), especially with stock cooler…My Asus 7970 runs at 102degress on each core 24/7, been doing it for months.
-
Im also interested in knowing whats a normal temperature for a 7970 to operate on 24/7.
right now i have 2x7970s in a case(waiting for risers to arrive). What temperatures should i stay under ? -
[quote name=“TheRedViper” post=“39915” timestamp=“1386229761”]
Im also interested in knowing whats a normal temperature for a 7970 to operate on 24/7.
right now i have 2x7970s in a case(waiting for risers to arrive). What temperatures should i stay under ?
[/quote]my card is about 70-72 degree 24/7
-
Wow, I’m happy to have gotten so many replies. You rock, guys! :)
[quote name=“angloblaxon” post=“39836” timestamp=“1386213128”]
where are these fans positioned? they should be in front of the cards or slightly angled towards the fans. The side or back is a waste. if you have them in back of the cards it will actually cause the air to push against the cards exhaust. ( most cards take air from the case and push it out the back )
[/quote]I think I was too tired late last night… even though the temp went to 100 on one of the two 7990 GPU’s before I mounted the fans, there is definitely room for improvement. I will mount the fans so one is sucking in the exhaust end and one is blowing in the other end.
[quote author=RIPPEDDRAGON link=topic=5398.msg39851#msg39851 date=1386214608]
[quote author=HizzleFizzle link=topic=5398.msg39811#msg39811 date=1386208504]This is what my settings batch file looks like: cgminer.exe --scrypt -o stratum+tcp://pool.d2.cc:3333 -u X -p X --intensity 13 --shaders 2048 [/quote]
[b]These are not all of your settings. Cgminer has chosen a TC automatically for you and that could be bad…i have seen it before. Look for a file similar in name to this:
scrypt130302Tahitiglg2tc21105w256l4.bin
and post what your tc value is in the name. Mine would be a tc of 21105. Also read this and use the template to the best of your knowledge:
[url=http://forum.feathercoin.com/index.php/topic,5418.0.html]http://forum.feathercoin.com/index.php/topic,5418.0.html[/url][/b]
[/quote]
I experimented a bit with the TC last night. When I put it to 8192, it runs without any HW errors now (crazyinside, exactly as you just suggested :) ). But if I set it to the 21105 you mention, it gives the following message:
[b]Maximum buffer memory device 0 supports 536870912
Your scrypt settings come to 1383137280
Error -61: clCreateBuffer (padbuffer8), decrease TC or increase LG.
Failed to init GPU thread 0, disabling device 0[/b]
What did you set you LG to?
I only have 4GB RAM in the machine and I’m running Win7 32bit (didn’t realise when buying, that this could be important). Could this have something to say in the matter? Maybe I should put in more RAM and change to a 64-bit OS?Nice work on the template, Ripped! I will fill it out after writing this post.
[quote author=specter link=topic=5398.msg39878#msg39878 date=1386217786]
I would take it out of the plastic box and see what your temps do. sounds like your box is creating a pool of hot air to sit where a case as flow through design. Just because there are holes does not mean it will cool better. I would also try a box fan or something blowing down into the box, basically something to force cold air in.
[/quote]I might try that soon :)
[quote author=svennand link=topic=5398.msg39895#msg39895 date=1386222039]
And 7990 cards are notoriously hot (dual 7970 on a single pcb…), especially with stock cooler…
My Asus 7970 runs at 102degress on each core 24/7, been doing it for months.
[/quote]
Hmm… I’m starting to consider using my “right of withdrawal” and returning the 7990. It is not a cheap product, and I was hoping for something good. Right now I have lowered the second GPU to intensity 11 and still the temp is 95C. The first GPU went down 69C even though it’s still running at 13. I think this is more or less proof that construction of this card is bad. To me, it seems like all the heat from the first GPU is dissipating through the second GPU. They are probably on the same heat pipe.That is really not a clever design, if I am right.
[quote author=Pyxis link=topic=5398.msg39890#msg39890 date=1386219307]
Can you show us a picture of your rig?What is the ambient temperature where you have the rig? (and is it summer there/do you live in a desert etc :) )
[/quote]
This is it: https://dl.dropboxusercontent.com/u/3267314/Hot miner.JPGI live in Denmark, in the chilly nordics, so outside temperatures are getting low. I’m guessing the room is usually around 22C. I actually turned off the heating because the miner keeps the room at a decent temperature all by itself ;)
[quote author=MindaugasC link=topic=5398.msg39918#msg39918 date=1386230090]
my card is about 70-72 degree 24/7
[/quote]
I am very envious! ;) -
RIPPEDDRAGON’s template:
GPU (including manufacture): 1 x XFX R7970 DD AND 1 X ASUS 7990 (Malta chip) with 2 GPU’s on.
Mining program(s): CGminer 3.1.0
Mining Pool: d2
[b]GPU Core Clock: [/b]On 7970, 1050 MHz. On 7990, 1000 MHz.
[b]GPU Memory Clock: [/b]1500 MHz on all.
[b]Thread Concurrency: [/b]8192
[b]Average KH/s: [/b]Around 500
[b]The rest of your miner settings: [/b]–scrypt -o stratum+tcp://pool.d2.cc:3333 -u X -p X --intensity 13 --thread-concurrency 8192
[b]Any thing else you deem important: [/b]I just tried turning off the second GPU on the 7990 entirely. Its temperature is currently 96C while the first GPU is at 54C, so it is clear that the heat is going from one GPU into the other.
[b]The nature of your problem: [/b]The GPU’s are too hot and the hash rates are too low. I’m focusing on getting the temp down first and then I will focus on getting the hash rate up. I am 2 inches away from a phone call to the dealer to tell them that I’m returning that 7990. Unless you guys tell me otherwise, that is what I will do. -
A box fan should definitely help. I think you have to much heat buildup.
Add
[code]setx GPU_MAX_ALLOC_PERCENT 100[/code]To the top of your bat file. This might fix your high thread concurrency error
-
Thanks, angloblaxon.
Am I supposed to put that line in the batch file before or after the other stuff? After, right? If I put it before, the same thing happens, but if I put it after, other errors occur and it disables all the GPU’s.
[b]EDIT: Sorry… I read your post properly now. You said to put it in before :)
[/b]
Right now, I took it out of the plastic crate and it has made a difference. They are running at 90C (7970) and 84C and 88C (7990). There has been a small improvement for the 7990, but the 7970 is getting hotter! This is “great”! :PAnd here I thought I was going to be building an easy money making machine and be filthy rich in no time with no effort ;)
-
Hi again.
I figured out that “setx GPU_MAX_ALLOC_PERCENT 100” is just a line that I should type in from a command promt and then it should stay there from then on.
However, when I set the TC to 21105, I get the following when I launch CGminer:
error -4: Enqueueing kernel onto command queue. (clEnqueueNDRangeKernel)
error -4: Enqueueing kernel onto command queue. (clEnqueueNDRangeKernel)
GPU 0 failure, disabling!
GPU 2 failure, disabling!
GPU 1 failure, disabling!So I’m back to a TC of 8192.
Any clues? It would be nice to get those GPU’s going a bit faster.
By the way, for now, I am coping with the heat by leaving the door to my balcony open. It works just fine for now :)
-
[quote name=“HizzleFizzle” post=“40171” timestamp=“1386266859”]
Hi again.I figured out that “setx GPU_MAX_ALLOC_PERCENT 100” is just a line that I should type in from a command promt and then it should stay there from then on.
However, when I set the TC to 21105, I get the following when I launch CGminer:
error -4: Enqueueing kernel onto command queue. (clEnqueueNDRangeKernel)
error -4: Enqueueing kernel onto command queue. (clEnqueueNDRangeKernel)
GPU 0 failure, disabling!
GPU 2 failure, disabling!
GPU 1 failure, disabling!So I’m back to a TC of 8192.
Any clues? It would be nice to get those GPU’s going a bit faster.
By the way, for now, I am coping with the heat by leaving the door to my balcony open. It works just fine for now :)
[/quote]That tc value is a custom one for my MSI cards that i found, 8192 and 24000 are the two people use the most.
-
Alright, thanks, RD :)
BUT! I think I just fried my 7990!
I just tried some settings out that I found [url=https://litecointalk.org/index.php?topic=672.85;wap2]here[/url] from the user “Pagan”.
The 7970 had the same speeds, but my 7990 went to more than 700KH/s on each core until the machine turned off after a about 30 seconds.This is what I put the card through:
cgminer.exe --scrypt -o stratum+tcp://pool.d2.cc:3333 -u X -p X --intensity 13 --shaders 2048 --thread-concurrency 8192 -I 13 -g 2 -w 256.When I started the machine again, and started CGminer with my regular configuration, the GPU’s on the 7990 weren’t running. I then shut down the machine and removed the 7970 to focus on the 7990. When I try mining with my old configuration, it hangs after a while. I guess I must have fried it already :o
I saw no smoke, but there was a faint smell of burnt electronics.Oh the humanity!
What does -w do? I can’t find it in the readme.txt and guess it much be an abbreviation that the program still recognises. Was this what has affected my card perhaps?
I feel stupid…
EDIT: I guess it must be this one (found in the readme.txt):
[i]–worksize|-w Override detected optimal worksize - one value or comma separated list[/i]
But I still don’t understand what it does? -
Update!
Things are going better. Nothing seems to be fried anyway. Maybe it just needed to have the power disconnected for some time to “go back to zero”.
I did some moving around and right now I’m getting around 1900KH/s from those two cards. The hottest one is running at 90 - 92C.
I mounted three 80mm fans blowing right at the three fans of the 7990, which helps it move a lot more air through the vents.
I’m considering whether it might be safe to put it on my balcony during winter, and I would like to hear your opinions.
The balcony is closed off with a bunch of big windows. The windows keep out all rain and snow, but some wind gets through, so the air is always moving a bit. Because of the moving air, water doesn’t condense out there, so I’m thinking that it should be safe, and the GPU’s would have an easier time breathing. During winter, the temperature might go below zero, but as far as I know, most computer parts are designed to be able to handle that. Especially since air humidity goes down as the temperatures drop.I would really appreciate some views on the matter as I have no previous experiences with this and might be overlooking something important.
Thanks for helping out, guys! :)
-
You thought not to make liquid cooling? ::)
Suppose we remove coolers manufactures, insert copper rods and put these in oil or water, inverted plates and rods in cooling tanks. I’ve been immersed in oil systems, but I think the water or coolant is better …
I am seriously thinking about it for my mining equipment… ;)(sorry, I do not speak English, I use translator) ;)
-
You’re describing improperly seated/installed/sized heat sinks. Have you considered aftermarket cooling products, or maybe just re-seating the existing one with some new thermal paste?
There’s plenty of guides online on how to (and more importantly how NOT to) apply thermal paste correctly.
-
[quote name=“Ernesto” post=“40345” timestamp=“1386290978”]
You thought not to make liquid cooling? ::)
[/quote]
I thought about it, yes. But I prefer not to do that. Liquid cooling is a whole new thing to learn and right now I have enough to learn already. But thanks for the suggestion :)[quote author=Kevlar link=topic=5398.msg40349#msg40349 date=1386292014]
You’re describing improperly seated/installed/sized heat sinks. Have you considered aftermarket cooling products, or maybe just re-seating the existing one with some new thermal paste?There’s plenty of guides online on how to (and more importantly how NOT to) apply thermal paste correctly.
[/quote]You are absolutely right. I should have thought about that. I used to repair computers for a living, so I know how bad factories can be at using thermal paste. I will try that.
But still, I am interested to hear what you all think about putting the miner in the balcony. On top of lowering the temperature, I would also get the noise out of the living room.
The thing is, I am sharing a large apartment with 3 study buddies. They agreed that it would be okay for me to keep the miner in the living room, but the noise caught us all by surprise. I know the noise level would go down a bit with better thermal conductivity between heatsinks and hardware, but I think it would still be annoying to have next to you while watching TV.
So yeah… how about that balcony? :P